is also recognized by freshly isolated tumor infiltrating lymphocytes, p939 represents a good ... defined epitopes, some of which are unique to a given tumor,.
B ~ e f Det~nitlee
Report
Mass Spectrometric Identification of a Naturally Processed Melanoma Peptide Recognized by CD8 +
Cytotoxic T Lymphocytes By Chiara Castelli,* Walter J. Storkus,r MarkusJ. Maeurer,r Dina M. Martin,~ Eric C. Huang,S Birendra N. Pramanik,$ Tattanahalli L. Nagabhushan,$ Giorgio Parmiani,* and Michael T. Lotze~ From the "Istituto Nazionale Tumori, Milan, Italy; IDepartment of Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania 15261; and SMolecularSpectroscopyand Biotechnology, Schering Plough Research Institute, Kenilworth, New Jersey 07033
Sutnlmary We and others have previously reported that melanoma-specific, cytotoxic T lymphocytes (CTL) define a minimum of six class I-presented peptide epitopes common to most HLA-A2 + melanomas. Here we show that three of these peptide epitopes are coordinately recognized by a CTL clone obtained by limiting dilution from the peripheral blood of an HLA-A2 + melanoma patient. Tandem mass spectrometry was used to characterize and sequence one of these three naturally processed melanoma peptides. One of the potential forms of the deduced peptide sequence (XXTVXXGVX, X = I or L) matches positions 32-40 of the recently identified melanoma gene MART-1/Melan-A. This peptide (p939; ILTVILGVL) binds to HLA-A2 with an intermediateto-low affinity and is capable of sensitizing the HLA-A2 § T2 cell line to lysis by CTL lines and clones derived from five different melanoma patients. A relative high frequency of anti-p939specific effector cells appear to be present in situ in HLA-A2 + melanoma patients, since p939 is also recognized by freshly isolated tumor infiltrating lymphocytes, p939 represents a good candidate for the development of peptide-based immunotherapies for the treatment of patients with melanoma. uring the last several years, a number of studies have D been conducted on the autologous CD8 § T cellmediated response to human melanoma (1-17). The emerging picture indicates that melanomas express multiple T celldefined epitopes, some of which are unique to a given tumor, while others are shared by allogeneic, HLA-matched melanomas (2-4, 6-9, 11-17). These epitopes appear to represent short 9-10 amino acid peptides derived from tumor-associated antigens that are presented by M H C class I antigens to CD8 + T cells (5-9, 17, 18). While many class I alleles have been reported to represent restriction elements for tumorreactive CD8 § T cells (6, 7, 10), the HLA-A2.1 allele (expressed by 45% of melanoma patients) appears to play an immunodominant role in presenting melanoma epitopes (11). At least six different CD8 + T cell-defined epitopes appear to be expressed by multiple HLA-A2 + melanomas (12, 13, 19). The identification and sequencing of these individual epitopes should allow for the design and testing of peptidebased immunotherapies for the treatment of melanoma. Materials and Methods Cell Lines. Two melanoma cell lines have been used throughout this study. Me1624 (HLA-A2, -A3; ~B7, -B14; -Cw7: obtained from 363
the Surgery Research Branch of the National Cancer Institute, Bethesda, MD) and Me19742 (HLA-A2, -A24; -B13, -B18; -Cw6, -Cw7: obtained from the Istituto Nazionale Tumori, Milan, Italy) were cultured in RPMI 1640 media supplemented with 10% heatinactivated fetal bovine serum, 100 IU/ml penicillin, and 100 #g/ml streptomycin (all reagents from Gibco BRL, Gaithersburg, MD). The CTL clone A83 (autologous to Me19742) recognizes a common melanoma antigen and is restricted by HLA-A2 (16). The tumor infiltrating (TIL) 1235 line and TIL 501.A42 clone were kindly provided by Dr. Y. Kawakami (National Institutes of Health, Bethesda, MD) and have been previously reported to recognize p2735 of the MART-1 antigen presented in the context of HLA-A2 (17). TILs 5403 and 6970 were isolated from metastatic melanoma lesions of HLA-A2+ patients as previously described (20), cultured for 7 d in the presence of autologous tumor, and used directly as effector cells in cytotoxicity experiments. All TIL lines and clones were cultured in AIM-V media (Gibco BILL) supplemented with 10% heat-inactivated human AB serum (Gibco), and 300 IU/ml rhll~2 (Cetus Corp., EmeryviUe, CA). Acid Elution of MHC Class 1-presentedMelanoma Peptides. Acid elution of, and reverse-phase high performance liquid chromatography (HPLC) resolution of, melanoma peptides was performed as previouslydescribed (12). Individual HPLC fractions were lyophilized and reconstituted in 200 #1 of Hank's buffered saline (Gibco BRL) and stored at -20oC for use in cytolytic assays. Alterna-
j. Exp. Med. 9 The R.ockefeller University Press Volume 181 January 1995 363-368
9
0022-1007/95/01/0363/06 $2.00
tively, for mass spectrometric analyses, peptides were lyophilized and reconstituted in 50/~1of 50% acetonitrile/50% water and stored at -70~ until used. Reconstitutionof T Cell Epitopes. In peptide-pulsing assays, 10 /~1 of reconstituted peptides was added to microwells containing 104 SlCr-labeled T2 cells, 0.2 #g human B2-microglobulin (B2-m; Sigma Chemical Co., St. Louis, MO), and 0.2/~g MA2.1 (antiHLA-A2.1; American Type Culture Collection, lkockville, MD, reference 21) monoclonal antibody in a total of 125/d. Peptide loading was facilitated by the presence of the B2-m and MA2.1 reagents (22-24), and was allowed to proceed for 2 h at room temperature. Effector T cellswere then added at a 10:1 effector-to-target cell ratio (unless otherwise stated) and cytolytic assays performed as previously described (12). HLA-A2 StabilizationAssay. Variousconcentrations of synthetic peptides (as noted in text) were incubated for 18 h at room temperature with 106 T2 cells (25), I/~g B2-m, and 2 #g MA2.1 monoclonal antibody. Cells were then washed twice with buffered saline and stained with FITC-conjugated F(ab')2 goat anti-mouse Ig (Organon Teknika, Durham, NC) for 30 rain at 4~ After two additional washes with buffered saline, cells were fixed with 4% formalin (Fisher Scientific Co., Pittsburgh, PA) and flow cytometry performed on a FACScan| (Becton Dickinson, Mountain View, CA) as previously noted (22). Results are reported as percentage of increase (over control) of MA2.1 mean fluorescence channel number reactivity. Controls represent T2 cells cultured with ~2-m and MA2.1, in the absence of synthetic peptide. Mass SpectrometricAnalysis. Mel 9742 HPLC fractions 47 and 48 were pooled, lyophilized, then reconstituted in 50/~1 of 50% acetonitrile, 50% double distilled H20, and stored at -70~ A 10 #1 aliquot of this material was introduced into an API III tandem mass spectrometer (PE-Sciex, Ontario, Canada) via the articulated ion-spray interface. The sprayer needle was held at 4,500 V with a coaxial sheath of nebulizing gas (compressed air) flow. Profile mass spectrum was obtained for peptide samples by scanning the first quadrupole from mass-to-charge (m/z) 500 to 1,600 in 3.37 s. The final spectrum was averagedfrom 10 scans. A mass spectra/mass spectra (MS/MS) product ion spectrum was obtained for the p941 by scanning the fragment ions resulting from collision with Ar gas. The sequencewas assigned as XXGVXXTVX, where X = isoleucine or leucine, each with residue mass of 113. SyntheticPepticles. Peptideswere synthesizedusing FMOC chemistry by the Peptide Synthesis Facility (Shared Resource) of the Pittsburgh Cancer Institute. Each peptide was purified to >95% homogeneity by reverse-phase (RP) HPLC and the identity of each peptide confirmed by MS/MS. The following peptides were synthesized: p939/MART-1 32-40:ILTVILGVL; gpl00 280-288: YLEPGPVTA (9); HIV-nef 73-82:QVPLRPMTYK (26); influenza A matrix, Flu M1 58-66:GILGFVFTL (24); p53 186-196:DGLAPPQHLIR (22); and p53 264-272:LLGRNSFEV (22). Results
Specific Recognition of HLA-A2-presented Melanoma Peptides by CTL Clone A83. Melanoma 9742 (Md 9742) and mdanoma 624 (Mel 624) were acid treated (12) and peptides obtained from "~5 x 109 cell equivalents. These peptides were then resolved by P,P-HPLC (see Materials and Methods for details). Aliquots of individual fractions were pulsed onto the HLA-A2 + T2 cell line, in the presence of human/32-m and the anti-HLA-A2 monoclonal antibody MA2.1, and tested for recognition by antimdanoma CTL done A83 in 4-h cy364
totoxicity assays. Three bioactive peaks were identified for autologous Mel 9742 peptides, resolving in HPLC fractions 42-43, 45, and 47-48 (Fig. 1). A similar pattern of bioactive peaks was identified for allogeneic HLA-A2 + Mel 624 fractionated peptides (Fig. 1). Peak 1 (HPLC fraction 42-43) and peak 3 (HPLC fractions 47-48) exhibited comparable efficacy in sensitizing the T2 target cell to lysis by CTL clone A83, with peak 2 (HPLC fraction 45) displaying a somewhat lesser capacity to do so. HLA-A2 restricted, melanoma-specific TIL 1235 and CTL clone A42 also recognized these same A83identified bioactive peaks as detected by cytolysis of T2 peptidepulsed target cells, with peak 2 reactivity approximating that of peaks 1 and 3 (Fig. 1). Mass SpectrometricAnalysis. Me19742 HPLC fractions 47 and 48 (peak 3) were pooled and analyzed using tandem mass spectroscopy, 10/~1 of fractionated material, corresponding to 20% of the pooled fractions and representing 109 cell equivalents, were injected into the electrospray ionizing source and the MS of the peptides recorded on a triple quadrupole mass spectrometer (PE-Sciex). The summation of mass spectra for peptides with m/z = 500-1,600 is reported in Fig. 2 A. At least 39 different peptides were detected, with one peptide (m/z = 941) representing the predominant species. CoUision-induced dissociation was performed on this ion yielding the daughter ion spectrum shown in Fig. 2 B. The spectra was interpreted as defining a nine-amino acid peptide of Mr - 939 with the sequence XXXTVXXGVX, where X = isoleucine or leucine (single letter amino acid designations). Since isoleucine and leucine exhibit identical masses (Mr -- 113), 32 potential variants of this peptide sequence are possible. A search of the GenBank database yielded a complete homology for one of these sequences, ILTVILGVL, with a recently cloned melanoma-associated gene Melan-A (accession number HSU06654, 15) also known as MART-1 (14). p939 Binds to HLA-A2 with Intermediate-to-LowA.~nity. The p939 sequence ILTVILGVL was synthesized and shown to elute in HPLC fractions 48/49, approximating the naturally processed bioactive form eluted from Mal 9742. This peptide was then analyzed for its ability to bind to HLA-A2 using an HLA-A2-specific stabilization assay implementing the T2 cell line (22, 27). The T2 cell line exhibits severely depressed cell surface class I expression resulting from a genetic lesion affecting peptide transport into the endoplasmic reticulum (25). T2 expression of class I (in particular HLA-A2) at the cell surface can be enhanced by incubation at reduced temperature in the presence of exogenous 3a-m and peptides capable of binding to the HLA-A2 allele (22-24). We incubated a panel of synthetic peptides, induding p939, with the T2 cell line at room temperature (23~ for 18 h and evaluated their capacity to stabilize HLA-A2 expression on the cell surface of T2 cells. Those peptides binding to HLAA2 yielded an elevated reactivity with the MA2.1 (anti-HLAA2) monodonal antibody compared with non-peptide-treated or irrelevant peptide-treated T2 cells. By performing a peptide dose titration analysis, a qualitative assessment of the relative binding capacity of each peptide for HLA-A2 was obtained (22). Analysis of a series of peptides revealed that
Identificationof a Natural Melanoma Epitope
Clone A83
TIL 1235
1o
io
no.
20
Clone A42
tO
' 038 " 40
42
44
46
48
SO
52
1 38 " 4 ; " 41 " 4~1 " 4 ; " 48 " 5; " S
03'8
4 ; " 42 - ~
" 4'6 " 4? " !3 = 5'2
HPLC Fraction Number
the Flu M1 58-66 and p53 274-282 peptides were quite effective in stabilizing HLA-A2 expression on the T2 cell line (half-maximal activity at '~5 riM), while the p939 peptide was less effective (half-maximal activity at '~200 nM) (Fig. 3). The efficacy of the p939 peptide was comparable to that observed for the gpl00 280-288 peptide, recently identified by Cox et al. (9) as a melanoma-assodated, HLA-A2-presented T cell epitope. The gpl00 280-288 peptide was similarly reported to bind HLA-A2 with intermediate-to-low affinity (9). As controls, the p53 186-196 stabilized HLA-A2 very poorly and the HLA-A3 binding HIV-nef 73-82 peptide did not stabilize HLA-A2, as previously reported (22). Reconstitution of a T Cell Epitope by p939. Epitope recon-
.ooj
Q
stitution experiments using p939 were performed and sensitization of T2 cells to lysis by antimelanoma CTL was evaluated over a wide range of peptide concentrations (250 riM-2.5 pM) (Fig. 4). Four different HLA-A2 restricted, antimelanoma CTL were evaluated in these studies: the bulk cultured TIL 1235 line, two CTL clones (A83 and A42), and the fresh TIL 6970 line. Despite the apparent low affinity of p939 for HLA-A2, each of these CTL recognized peptide-pulsed T2 targets. Interestingly, while half-maximal sensitization for lysis mediated by TIL 1235, clone A83, and TIL 6970 were reached at ",~10pM of peptide, the A42 clone required ~3,000 pM of peptide for half-maximal sensitization. TIL 1235 and CTL clone A42 recognition of p939-pulsed T2 targets also
94t
(A)
~o
600
800
1000
1200
1400
m/z
11._~4 2 2 7
100 §
~
7,~
~
50
(B)
X
328
427
540
653
710
809
T
V
X
X
G
V
X
231
130
X
939 e 2 5 7 1 2 6 1 3 5 0 _ _ 0 0 3 8 7
86
100
Figure 1. Melanoma TIL lines and clones recognize three predominant epitopes resolved by RPHPI.C. MHC class I-presented peptides derived from Me19742 or Mel 624 were extracted and fractionated by RP-HPLC. After lyophilizing, each fraction was reconstituted in 200 #1 of buffered saline and 10/~1 used to pulse onto SlCr-labeled T2 target cells in the presence of human ~2-m and the anti-HLA-A2 monoclonal antibody MA2.1 (as described in Materials and Methods). CTL were then added at an effector-totarget cell ratio of 2:1 (CloneA83) or 10:1 (TIL 1235 and CloneA42) in a standard 4-h cytotoxicity assay. HPLC fractions 1-38 pulsed onto the T2 cell line were not recognized by CTL.
~
939
OH-
5O
~00
300
400
~
r
700
8(10
m/z
365
Castelli et al.
Brief Definitive Report
900
1000
Figure 2. Mass spectrometric analysis of Me19742 fraction 47/48 peptides. HPLC fractions 47 and 48 derived from 5 x 109 melanoma 9742 cells were combined and 10 #1 corresponding to one-fifth of the original volume was injected into the dectrospray ionization source of a triple quadrupole mass spectrometer (PE-Sdex). Summation of the mass spectra obtained in the 500-1,600 m/z range is depicted in A. The (M+H) + ion at m/z = 941 was selected for fragmentation to generate sequence data and used to generate a similar CID spectrum (shown in B). Predicted (and actual, underlined) masses for the fragments of the type b (carboxyl terminal cleavage)are shown above and type y (amino terminal cleavage) below the deduced sequence XXTVXXGVX (Mr = 939), where X = isoleucine or leucine.
/f
]tOO-
'~,
80-
.f.~ 80
Ii 2O
~
~;~ .... ;;2 .... ;;3"
~o~
~;s
Peptlde Concentration (nM)
Figure 3. p939 binds to HLA-A2 with moderate-to-low affinity. The stabilization of HLA-A2 on the T2 cell line in the presence of exogenous peptide was evaluated as outlined in Materials and Methods using the indicated synthetic peptides. HLA-A2 expression was evaluated using the MA2.1 (anti-HLA-A2 monoclonal antibody) in indirect immunofluorescence assays monitored by flow cytometry. Data are reported as percent increase in MA2.1 reactivity relative to T2 incubated in the absence of peptide. A, p53-1; El, Flu M1; O, gpl00; A, p939; I-1, p53-2; @, HIV-nef.
resulted in peptide-spedfic releaseof interferon-3, at p939 doses as low as 1 pM (data not shown). An additional HLA-A2restricted fresh melanoma TIL 5403 was also able to recognize and lyse T2 target cells pulsed with p939 peptide (data not shown). CTL recognition of T2 presented p939 could be inhibited by addition of the anti-HLA-A2 monoclonal antibody CRll-351 (data not shown). Discussion
We have identified a naturally processed, melanomaassociated epitope recognized by five distinct HLA-A2-
restricted, tumor-specific CTL lines and clones. This peptide was isolated from melanoma cell surface HLA-A2 class I complexes by mild add elution. The sequence of this peptide (p939: MART-1/Melan-A 32-40; ILTVILGVL) was deduced by mass spectrometry and a GenBank database search for homology, representing the second successful use of this technology to identify a tumor-associated epitope (9). Interestingly, the p939 epitope represents one of three epitopes coordinately recognized by two distinct antimelanoma CTL clones (A83 and A42) derived from two different HLAA2 + patients (14, 16). The other two epitopes may be represented by MART-l-derived nonameric peptides MART-1 27-35;AAGIGILTV and MART-1 29-37;GIGILTVIL, since the CTL clone A42 and TIL 1235 used in this study have been recently reported to react with these sequences(17). These latter two sequences have not been proven to be naturally processed and presented by melanoma. If they represent naturaUy processed and HLA-A2-presented epitopes, this would support the recognition of three distinct and overlapping peptide sequences (all share the MART-1 32-35;ILTV sequence) derived from a single protein precursor (MART-l) by two independent CTL clones. The high avidity of clone A83 for p939 is somewhat surprising since clone A83 is an autoreactive CTL; cross-reacting with melanoma and with normal melanocytes (16). If antitumor reactivity represents autoreactivity in human melanoma, the role of positive and negative selection in maintaining such clonotypic T cells in vivo is of significant interest. The "differential avidity" model (28) could account for the positive selection of clones such as A83. In this model, the critical parameter is the number of TCR engaged with peptide-MHC complexes on APC. When this number is moderate, positive selection may occur. The combination of a high affinity TCR with a specific peptide epitope displaying a low
80
60
C l o n e A83
TIL 1235
40 40-
20'
20-
...J.
0
0 40
TIL 6970
#
80 20
20
10 10
......
9
......
10-1 10 0
!
.....
101
I
......
10 2
i
.....
-i
......
10 3 10 4
i
....
10 5 1 6 ~
" 1 . .10 . . .0. . . 101 ' . . . .10 . . .2. . . .10 . . .3. . . 10 . . . 4. . . . 10 . . 5 10" 6
Peptide Dose (pM) 366
Identification of a Natural Melanoma Epitope
Figure 4. Epitope reconstitution with p939. Melanoma-specificCTL (cultured TIL 1235, clone A83, clone A42, and fresh TIL 6970) were assayed in 4-h cytotoxicity assays at effector-to-target ratios of 10:1 against stCr-labeled T2 cells pulsed with p939 at the indicated concentrations. Lysisof T2 cells not pulsed with p939 was
